In many wireless communications systems, such as the Universal Mobile Telecommunications Systems (UMTS), the process of allocating resources to a user entails extensive signalling and negotiation between the user terminal and different network nodes. These procedures are time consuming, adding considerable delay to the service setup times.
Further, in order to use network resources as efficiently as possible, different states are typically defined for the terminals in wireless networks. A terminal that is not currently engaged in a communication session can assume a state in which the use of power in the terminal and network resources such as memory and power is reduced. In such a state, the need for signalling between the terminal and network, that arises, for example because of mobility is also reduced. When a communication session is initiated the terminal's state must be changed, which is time consuming and poses a delay in the session establishment.
In fixed broadband networks, such as Ericsson's Public Ethernet solution, a different approach is taken. There, the network keeps network resources pre-allocated to the user, even before the user requests a service to be initiated.
The advantage of the latter approach is that it removes the need for time consuming resource allocation signalling when a communication session is to be set up. Therefore, the response time is shorter from service invocation until the data transfer starts. The drawback of this approach is that resources, such as buffers, memory for storing user contexts, etc., are allocated both in the network and in the user terminals, even when no communication takes place.